OpenAI accused of monitoring prompts and pressuring mathematician off Millennium Prize Problem credit

anshulkundaje · x · 2026-09-08

Mathematicians Leven and Tristan spent months working with various AIs on a Millennium Prize Problem; per ThomWolf, OpenAI learned of it, pushed its latest models along the same direction, then tried to control communication of the results and drop Leven from authorship.

@wildtypehuman argues this is bad for both OpenAI and Anthropic: prompts may be monitored for scientific/commercial value, and Anthropic is increasingly seen as competing with its own customers. With pharma companies disclosing drug-target IP to Claude, he expects many to stop using these tools or add strict internal guardrails.

Related event: OpenAI's Navier-Stokes attribution dispute draws wave of criticism(9 posts)→

Original post →

More from AGI Musings

AGI Musings channel →